摘要：A graphical tool for choosing the number of nodes for a neural network is introduced. The idea is to fit the neural network with a range of numbers of nodes at first, and then generate a jump plot using a transformation of the meansquare errors of the resulting residuals. A theorem is proven to show that the jumpplot will select several candidate numbers of nodes among which one is the true number of nodes. Then a single node only test, which has been theoreticallyjustified, is used to rule out erroneous candidates. The method has a soundtheoretical background, yields good results on simulated datasets, and shows wideapplicability to datasets from real research.
